Aarhus University Seal / Aarhus Universitets segl

No. 91: Road-killed larger animals in Denmark 2003-2012

Elmeros, M., Andersen, P.N., Sunde, P. Haugaard, L., Skov, Fl. & Madsen, A.B. 2014. Påkørte større vilde dyr i Danmark 2003-2012. Aarhus Universitet, DCE – Nationalt Center for Miljø og Energi, 82 s. - Videnskabelig rapport fra DCE - Nationalt Center for Miljø og Energi nr. 91. http://dce2.au.dk/pub/SR91.pdf 


Collisions between vehicles and large wildlife species may constitute a safety and economical problem for the drivers and a conservation problem for rare species. The need for knowledge and methods to manage this wildlife-vehicle conflict is increasing as deer populations increase. The main objectives have been a precise mapping of road-killed larger animals in Denmark and to identify spatial and temporal wildlife-vehicle patterns in order to reduce the number of road kills. Tracker dog handlers in the national Schweiss Tracker Dog Register and employees from the Nature Agency collected information on: coordinates for the collision site, time of collision and age and sex on medium sized and large wildlife species.

Data were primarily collected on the game species roe deer (Capreolus capreolus), red deer (Cervus elephus), sika deer (Cervus nippon), fallow deer (Dama dama) and red fox (Vulpes vulpes). Information on wildlife vehicle collision sites has been collected in 2003 - 2012. The data for the period 01/01/2003 – 31/10/2006 were reported by Andersen & Madsen (2007).

A total of 33,605 collision sites between medium-sized and large wildlife and vehicles were recorded from 01/01/2003 to 31 December 2012 comprising 29,695 roe deer, 1,140 red deer, 1,444 fallow deer, 141 sika deer 590 red foxes and 595 other species (e.g. badger, Meles meles, wild boar, Sus scrofa). The annual number of recorded wildlife-vehicle collisions increased from ca. 2,800 to more than 4,100. This increase cannot be explained by increased traffic loads or focus on the wildlife-vehicle conflict, but is attributed to increasing deer populations as indicated by the national game bag records.

Roe deer-vehicle collisions peak in May and in October-December (Figur 3.2). A similar pattern is seen for red deer collisions while fallow deer collisions only peak in the autumn months. 64 % of the recorded roe deer-vehicle collisions involved females reflecting a female-bias population structure due to selective hunting of males. Only in May similar numbers of males and females were recorded. For red deer and fallow deer males made up 55 % and 60 % of the recorded road kills.

Each Spring and Autumn it has been claimed that the switch between standard Central European Time (CET = UTC +1h) and Central European Summer Time (CEST = UTC+2h) the last Sunday in March and the last Sunday in October results in significant increases in wildlife-vehicle collisions, particularly during the morning hours. Comparing weekly numbers of roe deer-vehicle collisions before and after the changes in human time setting, no significant change in collision numbers was detected following the switch to CEST in Spring or to CET in Autumn. During spring, the numbers of collisions increased considerably from 3 to 7 weeks after the switch to CEST, i.e. in late April and early May (Figur 3.8). During Autumn, the weekly numbers of roe deer-vehicle collisions increased gradually during the eight weeks before the switch to CET, i.e. in September and October, after which the collision numbers stabilized for the next 6-7 weeks. Collision numbers drop following the switch to CET in Autumn.

Roe deer-vehicle collisions were analyzed with generalized linear modelling as resource selection model to identify important variables for the observed spatial and temporal collision patterns and model collision risk for each road stretch. The number of roe deer collisions on each 100 m road stretch in rural areas in the whole country were used as the response variable. The response variable was analyzed with road, traffic and landscape parameters. Landscape parameters were assessed on a 10x10 km, 1x1 km and 100x100 m square levels.

The frequency of roe deer collisions is highest on major roads (>6m wide) and expressways, and increased with increasing traffic intensities up to AADT at ca. 8,000 vehicles. At higher AADT the collision frequency declines. At the 10x10km-sq level the most important landscape variables were the Shannon-Weiner diversity index. At the 1x1km-sq level the most important variables were area with forest and area of habitat types protected by the Nature Protection Act. At the 100x100m-sq level the most important variables were length of forest edge. Roe deer collision frequency was positively correlated with all the landscape variables. Based in the statistical model the expected roe deer collision frequency was estimated for all road stretches in Denmark.

The best statistical model to explain temporal roe deer-vehicle collision pattern through the year include the variables: Month linked to Day-type (weekend or working day), Weekday, Hunting season and Lunar size. The frequency of roe deer-collisions is high on working days and during the hunting season. The roe deer-vehicle collision risk is approximately 10 % higher in periods around full moon than in periods with new moon.

Furthermore, the report reviews published evaluations of the efficiency of various mitigating measures to guide national and local road and nature authorities as to which mitigating measures have documented effect in reducing wildlife-vehicle collisions.

We outline projects to disseminate the information on high-risk road stretches to drivers and to collect additional information on wildlife mortalities on railways. We recommend that reliable data on numbers of road-killings of smaller and non-game species should be collected to monitor and assess the potential effects of incidental killings on the species’ population status.